单项专利价值的评估与定量评估指标体系的构建——基于邻域粗糙集与果蝇优化神经网络的单项专利价值评估  

Evaluation of a Patent and Construction of Quantitative Evaluation System——Based on Neighborhood Rough Set and Fruit Fly Algorithm Optimized Neural Network

作　　者：慎金花[1] 刘玥[2] 张更平[1] Shen Jinhua;Liu Yue;Zhang Gengping

机构地区：[1]同济大学图书馆,上海200092 [2]同济大学经济与管理学院,上海200092

出　　处：《大学图书馆学报》2020年第3期48-56,64,共10页Journal of Academic Libraries

基　　金：国家社会科学基金项目“需求和能力导向的大学图书馆专利情报服务机制研究”(批准号:15BTQ027)的研究成果之一。

摘　　要：为提高专利价值预测的准确性,促进专利权的转让,围绕如何以专利权转移过程中的专利定价客观评价专利价值,构建了针对单项专利可定量评估的指标体系,克服了专利价值评估中的主观性,使用邻域粗糙集方法排除冗余属性,进行特征选择,采用果蝇算法优化BP神经网络,降低了BP神经网络容易陷入局部极小的风险。通过实证研究发现,构建的专利价值指标体系可对单项专利进行定量评价,果蝇算法优化后的BP神经网络具有比较快速和准确的预测能力,在实际预测中具有良好的泛化能力和有效性。In order to improve the accuracy of patent value prediction and promote the transfer of patent rights,this study focuses on how to evaluate the patent value objectively in the process of patent transfer,and then constructs an index system for quantitative evaluation of a patent,in hope of overcoming the subjectivity in the evaluation of patent value.It uses the neighborhood rough set method to eliminate redundant attributes and the Fruit Fly Optimization Algorithm to optimize the BP neural network,in order to reduce the risk of that the BP neural network is easy to fall into local minimum.Through empirical research,it is found that the patent value index system constructed in this paper can quantitatively evaluate a single patent.The BP neural network optimized by Fruit Fly Optimization Algorithm has relatively fast and accurate prediction ability,together with good generalization ability and effective in actual prediction.

关 键 词：专利价值评估 邻域粗糙集 果蝇算法 BP神经网络 

分 类 号：G255.53[文化科学—图书馆学]